https://ph02.tci-thaijo.org/index.php/TJOR/issue/feedThai Journal of Operations Research : TJOR2024-12-18T14:29:40+07:00Associate Prof Dr.Kannapha Amaruchkulorjournal.th@gmail.comOpen Journal Systems<p>วารสารไทยการวิจัยดำเนินงาน (Thai Journal of Operations Research : TJOR) เกิดขึ้นจากความร่วมมือของคณาจารย์ และนักวิจัยในเครือข่ายการวิจัยดำเนินงาน (Operations Research Network of Thailand, OR-NET) โดยมีวัตถุประสงค์เพื่อส่งเสริมและเผยแพร่ผลงานทางวิชาการด้านการวิจัยดำเนินงานที่มีคุณภาพ วารสารไทยการวิจัยดำเนินงานเป็นวารสารอิเล็กทรอนิกส์ (E-Journal) ที่มีกำหนดออกปีละ 2 ฉบับ คือประมาณเดือนมิถุนายน และเดือนธันวาคมของทุกปี </p> <ul> <li class="show">วารสารไทยการวิจัยดำเนินงาน (Thai Journal of Operations Research) <strong>ได้รับการจัดกลุ่มวารสารที่ผ่านการรับรองคุณภาพของ </strong><strong>TCI อยู่ในวารสารกลุ่มที่ 1</strong></li> <li class="show"><strong>ไม่มีค่าใช้จ่ายในการตีพิมพ์</strong></li> <li class="show"><strong>จากประวัติที่ผ่านมาใช้เวลาในการดำเนินการไม่เกิน 3 เดือน/บทความ</strong></li> </ul>https://ph02.tci-thaijo.org/index.php/TJOR/article/view/254616Multiple Objectives Optimization: from development in the past to future of applications2024-06-28T17:02:44+07:00Pakorn Adulbhanathakorn.kengpol@gmail.comMario T. Tabucanonathakorn@kmutnb.ac.thAthakorn Kengpolathakorn.kengpol@gmail.com<p>Multiple objective optimization (MOO) is critical in operations research and decision-making. It has crucial benefits for industrial work regarding efficient resource management, such as raw materials, labor, machinery, time, finances, and production methods. This article will present the concept of the model. and the process of multiple objective optimization. The objective is to enable readers to understand techniques for using multiple objective optimization theory. The article reviews the literature and gives examples of case studies using mathematical models to help with multiple objective decision-making in many industrial sectors, such as resource planning process improvement and transportation management. The knowledge gained from this article is valuable and helps the readers apply the knowledge of multiple objective optimization to other types of problems.</p>2024-12-18T00:00:00+07:00Copyright (c) 2024 https://ph02.tci-thaijo.org/index.php/TJOR/article/view/253052An Integer Programming Model for Economic Crop Cultivation Planning with Single and Multiple Harvests in the Dry Season2024-04-11T16:49:31+07:00Apisuda Glinsukhonapisuda.gli@dome.tu.ac.thSuphanida Khongpensuksuphanida.kon@dome.tu.ac.thSupanun Angchuansupanun.angc@dome.tu.ac.thAua-aree Boonpermaua-aree@mathstat.sci.tu.ac.th<p>Water scarcity during dry seasons presents a significant challenge for farmers, forcing them to make crucial decisions regarding crop selection. Short-term economic crops, offering single or multiple harvests, are often favored due to their faster turnaround. However, each crop has unique cultivation and harvesting windows. Therefore, optimizing crop planning is essential to maximize farmer profit under constraints of water availability, budget, and time. This study proposes an integer programming model for optimized crop planning. The model considers cultivation area, budgetary limitations, restricted water access, and harvesting periods to achieve maximum profit for farmers. Real data from the Baan Lung Khian Soei Som, Phetchaburi, Thailand Sufficient Economy Learning Center was used to test the model. Python, along with the Gurobi Optimization Package, facilitated the analysis. Budget variations between 75,000 and 275,000 Baht, and water volumes ranging from 3,500 to 5,000 cubic meters, were considered during model testing. The results identified an optimal budget of 250,000 Baht for crop cultivation, leading to maximum profit. Additionally, an analysis of return on investment revealed that a budget of 100,000 Baht paired with a water volume of 5,000 cubic meters yielded the highest return, translating to a 132.37% profit margin. The findings of this study highlight the crucial role of effective planning and budget allocation in crop cultivation. Strategic planning under water, land, budget, and harvesting time constraints can significantly enhance farmers' profits.</p>2024-12-18T00:00:00+07:00Copyright (c) 2024 https://ph02.tci-thaijo.org/index.php/TJOR/article/view/253056Planning a Tourist Route with the Objective of Maximizing Tourist Satisfaction Within Budget and Time Constraints Using a Mixed Integer Linear Programming Model2024-04-11T17:10:58+07:00Teeradech Laisupannawongteeradech.lai@gmail.com<p>Planning a tourist route to create an optimal tourist itinerary is crucial because it can save travel time, travel expenses, as well as provide satisfaction and memorable experiences from the journey. This paper presents a mixed integer linear programming model for planning a tourist route that is based on the traveling salesman problem. The proposed model has two objectives. The first is the main objective which is to maximize the satisfaction of tourists with the tourist attractions on the tourist route. The second is the secondary objective which is to minimize the overall travel distance. The constraints considered in this study include ensuring that the expenses and time spent visiting all tourist attractions within the tourist route do not exceed the limits set by the tourists. Furthermore, the tourists must start at the starting point (meeting point or accommodation), go through all tourist attractions along the tourist route, and then come back to the starting point. In this paper, some tourist attractions may not be selected to be on the tourist route due to budget and time constraints. The proposed model was tested with three test problems (Problems 1 – 3) using data on six interesting tourist attractions in Bangkok, 11 in Nakhon Pathom, and 16 in Ayutthaya, respectively. Each problem was assigned a maximum visit time to all tourist attractions on the route, not exceeding 8 hours, and a budget for the entire tour, not exceeding 1400, 1800, and 500 Baht for Problems 1 – 3, respectively. The results showed that the proposed model could find an optimal tourist route for all test problems, where the number of selected tourist attractions on the optimal tourist route in Problems 1 – 3 are 6, 9, and 13, respectively. The solution from the proposed model can be used to create a travel timetable specifying arrival and departure times for each destination. Moreover, it can be used as an option to find an optimal tourist route for travelers in a real-life situation.</p>2024-12-18T00:00:00+07:00Copyright (c) 2024 https://ph02.tci-thaijo.org/index.php/TJOR/article/view/253055Comparative Study of Deep Transfer Learning Models and Data Augmentation Techniques for Real and AI-generated Portraits Classification2024-06-14T09:11:28+07:00Pailin Manaohwanpongsan.pr@ku.thSorachai Phetkhunthotpongsan.pr@ku.thJulalak Kaewwangsakoonpongsan.pr@ku.thPongsan Prakitsripongsan.p@gmail.com<p>Distinguishing between a human portrait and one generated by artificial intelligence (AI) is becoming increasingly difficult. As AI technology advances, making AI-generated portraits more similar to real people portraits, posing a significant challenge for classification systems. Portraits contain a vast amount of information regarding color, texture, lighting, and subtle details. Deep learning models, with their layered architecture, can effectively learn patterns and relationships within this data. This paper explores the power of transfer learning with CNNs and data augmentation to enhance accuracy for classifying real and AI-generated portraits. We leverage three pre-trained models (MobileNetV2, ResNet50, and EfficientNetV2S) on a dataset of 3,000 images (1,500 per class). The performance is evaluated with and without image augmentation, providing valuable insights into their combined effect. Our findings suggest that EfficientNetV2S without data augmentation achieved the highest accuracy of 94.67%. and 94.47% for F1-Score.</p>2024-12-18T00:00:00+07:00Copyright (c) 2024 https://ph02.tci-thaijo.org/index.php/TJOR/article/view/253060Analysis and Development of Transportation Approaches to Reduce Logistics Costs Using the Large Neighborhood Search Algorithm2024-05-13T06:58:09+07:00Waroonon Booncharoenwaroonon.bo@ku.thChansiri Singhtaunwaroonon.bo@ku.thRoongrat Pisuchpenwaroonon.bo@ku.th<p>This research focuses on analyzing and developing a transportation management approach to</p> <p>reduce logistics costs for a logistics company. The company is responsible for transporting automotive components from component manufacturers to automotive assembly plants. The proposed transportation management approach addresses the Heterogeneous Fleet Capacitated Vehicle Routing Problem with Pickup and Delivery (HFCVRPPD). Three approaches were considered. Approach 1 involves grouping pickup points with common delivery points and synchronized pickup times. Approach 2 entails grouping pickup points with common delivery points and closely scheduled pickup times. Approach 3 involves grouping pickup points with proximate delivery points and closely scheduled pickup times. The Large Neighborhood Search (LNS) algorithm is utilized to find solutions. The findings reveal that the third approach provides the most significant reduction in logistics costs for the company, achieving a 58.52% reduction. Additionally, it improves the average vehicle capacity utilization efficiency from the original 58.07% to 92.60%.</p>2024-12-18T00:00:00+07:00Copyright (c) 2024 https://ph02.tci-thaijo.org/index.php/TJOR/article/view/253366An efficient determination of order batching and design of picking sequence towards order-picking distance in a warehouse2024-05-23T10:29:40+07:00Arisara Pewnuanarisara.pe17@gmail.comSurat KongsomAkkaranan.pon@nida.ac.thAkkaranan Pongsathornwiwatakkaranan.pon@nida.ac.th<p style="font-weight: 400;">A frozen meat warehouse for import-export businesses is experiencing high unproductive travel distances during picking processes. This directly increases warehouse management costs due to a single unit-load picking protocol. Therefore, a new design for batch picking and picking sequence is crucial for improving warehouse performance. This research develops a two-step approach to address this problem: 1) Improving Warehouse Layout: Two methods, Class-based Storage and the Storage Location Assignment Problem (SLAP), are applied to investigate a new warehouse layout. The results of the first step show that SLAP outperforms other methods, reducing travel distance by 30%. 2) Designing Order Size and Picking Sequence<strong>:</strong> After establishing an efficient layout, the next step is to design the order size and picking sequence to further reduce picking distance and test sensitivity. Two approaches are employed: the Roulette Wheel Simulation method and a Genetic Algorithm. Performance testing of picking operations reveals that a new order size of 6 SKUs is optimal for batch picking. The designed picking sequence reduces unproductive picking distance by up to 30% compared to the traditional picking protocol.</p>2024-12-18T00:00:00+07:00Copyright (c) 2024 https://ph02.tci-thaijo.org/index.php/TJOR/article/view/253542Monthly Volumes of Water Inflow into the Large Dam Reservoirs in Eastern Thailand Forecasting by The Cuckoo Search Optimization Enhanced Decomposition and Holt-Winters Techniques2024-05-01T14:11:30+07:00Pradthana Minsanpradthana_min@g.cmru.ac.thWatha Minsanwathaminsan@gmail.com<p>This study aims to evaluate the efficiency of models through the integration of Cuckoo Search Optimization (CS) with Holt-Winters (CS-HW) and Decomposition (CS-D) for forecasting monthly inflow volumes into large dam reservoirs in Eastern Thailand, covering a total of 6 dams. This is compared with the Grid Search of Holt-Winters (Grid-HW) and the Classic Decomposition Model (Classic-D) using a training dataset spanning 66 months, and employing the Mean Absolute Error (MAE) as the criterion for the lowest MAE to assess model performance. The findings indicate that both CS-HW and CS-D models outperform traditional models, with CS-D demonstrating a significantly lower MAE than Classic-D, while CS-HW shows marginally lower MAE values than Grid-HW across all dams.</p> <p>For the long-term forecasting of dam reservoir inflow volumes over a 24-month horizon using a test dataset, criteria including Root Mean Square Error (RMSE), MAE, and Symmetric Mean Absolute Percentage Error (sMAPE) were utilized to evaluate model performance. The results show that for different dams, specific models were chosen as best fitting: Khun Dan Prakan Chon Dam selected the CS with Additive Holt-Winters (CS-HW+), Khlong Siyat chose the CS with Multiplicative Holt-Winters (CS-HWx), Naruebodindrashinta Dam opted for the Box-Jenkins model, Bang Phra Dam for the CS with Multiplicative Decomposition (CS-Dx), Prasae Dam for the Grid Search of Multiplicative Holt-Winters (Grid-HWx), and Nong Pla Lai for the Grid Search of Additive Holt-Winters (Grid-HW+). The outcome of the 24-month forecasting demonstrates clear seasonal patterns in the reservoir inflow volumes to the dams.</p>2024-12-18T00:00:00+07:00Copyright (c) 2024